JP2015528173A - How to handle data using data relations - Google Patents

Abstract

Systems and methods can provide for tracing data using metadata. In one example, the method collects a text description of a first object that includes a word, generates a vector that represents the text description, assigns a first weight value to the word, and a second weight to the word. Associating an object space with the word including a value assignment and associating the object space with the first object.

Description

  Embodiments generally relate to traversing data relationships. More specifically, embodiments utilize data relationships to analyze context information about one or more objects to determine object attributes and relationships and to determine the attributes and relationships during data retrieval / retrieval. About using.

  Existing search applications may typically only return result sets based on keywords treated with equal importance, and such applications may not provide any contextual information about the results.

  As a result, the search may give non-optimal results from the user's perspective.

  Various advantages of embodiments of the present invention will become apparent to those skilled in the art by reading the following specification and the appended claims and by referring to the following drawings.

FIG. 6 is a block diagram of an example object space generation scheme based on an embodiment. 1 is a block diagram of a logical architecture according to an embodiment. FIG. 6 is a block diagram of an example object data structure according to some embodiments. 6 is a flowchart of an example method for associating an object with an object space, according to an embodiment. FIG. 3 is a block diagram of an example cluster data structure in accordance with some embodiments. 6 is a table of examples of relative attributes according to some embodiments. AC are block diagrams of a graphical user interface (GUI) according to an embodiment. 6 is a flowchart of an example method for tracing data according to an embodiment. FIG. 3 is a block diagram of an example processor in accordance with some embodiments. FIG. 1 is a block diagram of an example system according to an embodiment.

  FIG. 1 shows a plurality of objects 24 having a plurality of corresponding text descriptions 26 of the objects 24. Each text description 26 shown includes a word and / or phrase that describes the associated object 24. The object 24 may be any data element that can serve as a basis for efforts to trace data. Thus, the object 24 may represent, for example, items, locations, individuals, products and / or services that are the subject of commercial transactions, non-commercial data research efforts, social networking queries, and the like. In the illustrated example, one or more topics 28 are identified based on the text description 26. Here, the text description 26 and the topic 28 may be further used to obtain the association information 30.

  As will be discussed in more detail later, association information 30 includes the degree of association between each text description 26 and topic 28, the degree of association between each object 24 and topic 28, each text description 26 and The degree of association between the words in the text description 26 and the degree of association between the words in the text descriptions 26 and the topic 28 may be reflected. Association information 30 may be used to group objects 24 into clusters 32 of object space 34. Here, the object space 34 may be used to respond to and facilitate search requests. Of particular note, the use of association information 30 in the context of text description 26 allows the search to be formulated and / or guided by words that are not normally used to distinguish objects 24 from each other. It can be done. Thus, the illustrated scheme provides users with an unexpectedly high level of granularity and flexibility.

  FIG. 2 illustrates a logical architecture 36 (36a-36i) that can be used for both creating and following the object space. In the illustrated example, the topic generation module 36a identifies one or more topics based on a plurality of text descriptions of a corresponding plurality of objects. The topic generation module 36a may use a topic model to identify one or more topics. Further, the first association module 36b may determine a first degree of association between each of the text descriptions and the one or more topics to obtain first association information. As will be discussed in more detail later, the first association information may be configured as a first set of vectors corresponding to the plurality of text descriptions. Thus, each vector in the first set of vectors may hold a weight value, each weight value indicating the degree of association between the text description and a particular topic.

  The illustrated architecture 36 includes a second association module 36c that determines a second degree of association between each of the plurality of objects and the one or more topics to obtain second association information. Is also included. Similarly, the second association information may be configured as a second set of vectors corresponding to the plurality of objects, and each vector in the second set of vectors may hold a weight value. Good. In this example, each weight value may indicate the degree of association between the object and a particular topic.

  In one example, the architecture 36 determines a third association module that determines a third degree of association between each text description and one or more words in the text description to obtain third association information. 36d is also included. The third association module 36d may configure the third association information as a third set of vectors corresponding to the plurality of text descriptions, each vector in the third set of vectors being A weight value for identifying the occurrence frequency of one or a plurality of words may be held. Further, the fourth association module 36e may determine a fourth degree of association between each word and the one or more topics to obtain fourth association information. As will be discussed in more detail later, the fourth association module 36e may configure the fourth association information as a fourth set of vectors corresponding to the plurality of one or more words.

  The first association module 36b may obtain first association information that reflects the degree of association between the text description and the topic using the third and fourth association information. In this regard, the first association module 36b uses a probabilistic graphical model (PGM) to propagate fourth association information (eg, word-topic relationship) to the plurality of text descriptions in the first association information. You may use a graphical model). Further, the second association module 36c may use PGM to propagate the fourth association information to the plurality of objects in the second association information. Other approaches may be used to obtain the first and second association information. The illustrated architecture 36 includes a cluster for grouping the plurality of objects into clusters based on first association information from the first association module 36b and second association information from the second association module 36c. -It also includes a module 36f.

  The architecture 36 may be provided for following the object space according to a search request from the end user. More specifically, the illustrated architecture 36 includes a first result module 36g that generates a first set of results based on the requested search scope and object space. In one example, the object space includes a first object and a second object arranged according to a relative relationship based on attributes. As will be discussed in more detail later, the reconciliation module 36h includes a first reconciliation mechanism configured to adjust the scope of the requested search and a second reconfiguration configured to adjust the search results based on the attributes. A user interface having an adjustment mechanism may be generated. The illustrated architecture 36 also includes a second result module 36i that generates a second set of results based on user input received via one or more of the first adjustment mechanism and the second adjustment mechanism. Including. Here, the second set of results includes the second object.

  Turning now to FIG. 3, the object data structure 10 is shown. Data structure 10 may include one or more elements, which may include objects, metadata entries, words, and object space. As will be discussed in more detail later, each of these elements of the data structure may be analyzed alone or in combination with another to improve the way the related data is traced (eg, in data retrieval). Level (or layer).

  The illustrated data structure 10 includes an object 11. As already mentioned, the object 11 can be any data element that can be the basis for a data traversal effort (eg, data retrieval). Thus, in this example, the object 11 is a graphic novel “Dark Knight Returns” which may be purchased from an online vendor. Thus, the object 11 can represent the first level of data.

  The object 11 may be sold in electronic commerce by two or more online vendors. Each of these online vendors may associate a text description of object 11 to assist in the customer's purchase of the product. The text description (or metadata) associated with the object 11 may represent a second level of data.

  Furthermore, the text description may include one or more words. Here, words related to the text description of the object may represent third level data. Thus, in this example, the first vendor (eg, Amazon) may associate metadata 12 with object 11. The first vendor metadata 12 may include words 15-17 (ie, “comic”, “graphic”, “novel”) to describe the object 11. Similarly, a second vendor (eg, eBay) may associate metadata 13 with object 11. The second vendor metadata 13 includes the words 15, 16, 18, 19, 21 (ie, “superhero”, “comic”, “graphic”, “literature”, “book”) to describe the object 11. May be. A third vendor (eg, Burns and Noble) may associate metadata 14 with object 11. The third vendor's metadata 14 may include words 16, 18, 20, 21 (ie, “graphic”, “superhero”, “DC”, “graphic”, “book”) to describe the object 11. Good. As shown in FIG. 3, the vendor may use only one word (eg, “DC”) to describe the object 11 and use the same word to describe the object 11 without any other example. Or the vendor may use the same word more than once when describing the object 11 (eg, use of “graphic” by a third vendor).

  As will be discussed in more detail later, data about the object 11 (eg, metadata 12, word 15 etc.) may be used to parse context information about the object 11 to determine the attributes of the object 11. The attribute may then be used as a parameter in the search. Thus, in this example, the use of the word 18 “superhero” may be considered an attribute of the object 11 (ie, the object has some relationship to the superhero concept). Another example is the use of the word “date” in a wallet description (eg “I had a wallet to go on a date with a boyfriend”) or “beach” in a summer banana split truffle description There may be. The attribute may then be used as a parameter in the search, resulting in an object being returned. Of particular note is that “dates” may not be commonly used by wallet vendors to quantify wallets and / or to distinguish between wallets. Similarly, “beach” may not be typically used to quantify truffles.

  Further, data regarding the object 11 (eg, metadata 12, word 15 etc.) may be used to associate the object 11 with one or more clusters. Depending on the situation, a cluster may be considered a category, topic, group, community or any other grouping of objects that can be significant in following data related to the object 11. Thus, in this example, the object 11 may be associated with a cluster and / or topic 22 “graphic novel” and a second cluster and / or topic 23 “book”. Thus, the cluster may represent a fourth level of data.

  Turning now to FIG. 4, a method 40 for associating objects with clusters in object space is shown. In this example, the search application may associate an object such as object 11 (FIG. 3) with one or more topics such as topics 22 and 23 (FIG. 3). In this example, the object may be the graphic novel “Dark Knight Returns”.

  The method 40 may be performed on a machine or computer readable storage medium, such as a programmable logic array (PLA), such as random access memory (RAM), read only memory (ROM), programmable ROM (PROM), flash memory, etc. In configurable logic such as field programmable gate arrays (FPGAs), composite programmable logic devices (CPLDs), for example application specific integrated circuits (ASICs), complementary metal oxide semiconductors (CMOS) or It may be implemented in a fixed function logic hardware using circuit technology such as transistor-transistor logic (TTL) technology or as a collection of logic and / or firmware instructions stored in any combination thereof. For example, computer program code for performing the operations shown in method 40 includes an object oriented programming language such as C ++ and a conventional procedural programming language such as C programming language or similar programming language. It may be written in any combination of one or more programming languages. The method may begin at process block 42.

  At processing block 44, the search application may collect a text description such as metadata 12 (FIG. 3) from the first source. At processing block 46, the search application may parse the text description into one or more words. Thus, in the example of metadata 12 (FIG. 3), the application parses the metadata into three words (ie, “comics”, “graphics”, “novel”).

  In this example, the application may be configured to parse each word in the text description. However, other approaches may be used. In another example, an application parses and uses only certain words in a text description (eg, no other word, noun, adjective, etc.) or even extracts a keyword and makes it equivalent It may be configured to replace with metadata.

  At processing block 48, the application may utilize one or more words of the first source text description to generate a vector representing the text description. The vector may describe how many times a word or concept appears in the text description provided by the source and may be used to describe the text description. Thus, the vector may be part of the third association information generated by the third association module 36d (FIG. 2) already discussed. Thus, in the case of metadata 12 (FIG. 3), the first vector representing the metadata from the first vendor may be based on the words “comic”, “graphic”, and “novel” that each appear once.

  In the case of multiple text descriptions (from more than one source), the application may generate a vector for each text description associated with the object. In such a case (ie, for multiple vectors), each vector may be weighted. In one example, each vector may be equally weighted, and in another example, different weights may be assigned based on, for example, relevance to the object.

  As will be discussed in more detail later, vectors may be used to effectively determine the degree of association between words and objects. In one example, a first word that is mentioned twice in a description for an object will have a greater degree of association with the object than a second word that is mentioned once. Further, if the extension considers the word as an attribute of the object (as discussed above), the vector may be used to describe the degree of association between the attribute and the object.

  In processing block 50, the application may assign a weight value to each word in the vector. This weight value may represent the relevance or importance of the word within that vector. In one example, the words in the vector may be weighted using a probabilistic graphical model (PGM) to determine the weight of each word in the vector. Thus, for a vector associated with metadata 12 (FIG. 3), the first word “comic” may be assigned a first weight and the second word “graphic” may be assigned a second weight. The third word “Novel” may be assigned a third weight. If each word in the vector is equally weighted, then each of these words is given a third (1/3) weight. In the case of multiple vectors, the word weights in the first vector may not be the same as the same word weights in the second vector.

  As will be discussed in more detail later, the weight attributed to each word may be used to determine the degree of association between the word and its vector. As discussed above, words may relate to object attributes and vectors may relate to text descriptions of objects. Thus, the weight attributed to each word may represent the degree of association between the word and its associated text description and the degree of association between the attribute and the associated text description of the word. .

  At processing block 52, the application may associate one or more clusters with each word associated with the object. In one example, the search application may use a topic model (eg, Latent Dirichlet Allocation (LDA)) to analyze words and determine weight values. This weight value may represent the probability (ranging from 0 to 100 percent) that the word is associated with one or more clusters. Thus, in the case of metadata 12 (FIG. 3), using the three words “comics”, “graphics”, “novel”, the topic model may be, for example, the probability for the first cluster “book” in object space and the The probability for the second cluster “graphic novel” in object space may be returned. The probability that a word is associated with a cluster may also be viewed as the degree of association between that word and that cluster.

  At processing block 54, the application may associate one or more of the determined clusters with the object. In one example, an application may generate a topic vector using Bayes' theorem and each object's PGM. A topic vector may represent the probability that a cluster is associated with the object. Thus, in the case of metadata 12 (FIG. 3), the application is only 98% of the association between the category “graphic novel” and the object and only 2% of the association between the category “book” and the object. May find out. The probability of an object belonging to a topic may be calculated as follows.

P (object, topic) = Σ _document Σ _word P (object, document, word, topic)
P (object, document, word, topic) = P (object) P (document | object) P (word | document) P (topic | word)
At processing block 56, the application may calculate the overall probability of the object in the cluster. In one example, the overall probability of the object may be 1 / n. Here, n represents the number of objects in the cluster. If appropriate, this weight may be modified to give preference to certain objects. In process block 58, the process may end.

  The ordering and numbering of the blocks depicted in FIG. 4 is not intended to imply an order of operation, excluding other possibilities. Those skilled in the art will appreciate that the systems and methods described above can be subject to various modifications, variations and modifications.

Thus, these various calculations may be used to represent various relationships between various layers of data about the object. As discussed above, these relationships then
Degree of association between the text description and the topic (eg first association information);
Degree of association between the object and the topic (eg second association information);
Degree of association between word and topic (eg fourth association information);
Degree of association between topics and words;
Degree of association between words and text descriptions;
It may be used to determine various degrees of association, including (but not limited to) the degree of association between the text description and the object.

  Each of these relationships between the various data types may be considered random variables, which may be used to follow the data (eg, in a data search on the object). More specifically, as discussed in more detail later, these relationships may provide the user with several options when searching for data, each option being an object that the user may be interested in. May be used to focus search on the other side.

  For example, the degree of association between an object (based on a word found in the object's text description) and the attribute may be used to focus the search. For example, consider a consumer looking for a large backpack with many textbooks. The presence of the word “book” multiple times in the text description of the backpack can be reflected in the vector for the object. The vector can then be used to focus the user's search based on attributes (ie, look for a backpack to carry the book rather than hiking). Similarly, the degree of association between an object and a text description may also be used to focus the user's search. In this regard, “book” may not be a word typically used to quantify backpacks and / or distinguish between backpacks.

  Next, in another example, the degree of association between the first object and the second object may be used to focus the search. The degree of association between the first object and the second object is determined by, for example, the degree of association between an attribute and the first object and the degree of association between the attribute and the second object. May be. In one example, the degree of association between the first object and the second object may be used to generate a relative relationship (ie, similarity, dissimilarity, etc.). Here, the relative relationship may be calculated using a topic vector of each object related to a certain cluster, for example.

  Thus, for example, the frequent presence of the word “book” in the text description of the first backpack and the absence of the word “book” in the description of the second backpack can be attributed to the first backpack over the attribute “book”. A relative relationship with the second backpack may be generated (ie, the first backpack may carry books more than the second backpack). This relative relationship may be used in a “back to school” search for backpacks to be used by students.

  In yet another example, the degree of association between objects and clusters may allow search applications to organize and arrange the objects in the cluster. Thus, a search for a specific object for a specific attribute may in some cases begin with a search for a cluster. Once the right cluster is found, the desired attributes may be used to further focus the search to the desired object in the cluster. In this way, one or more objects in one or more clusters may be arranged to generate a map or graph of one or more dimensions or attributes.

  Thus, for example, taking the example described immediately above, a cluster of 10 backpacks may be organized in the object space according to their various attributes, for example using the topic vector of each object for that cluster . That is, a certain backpack (ie, object) may be organized within a cluster (ie, backpacks) using its relative relationship to other objects in space. Thus, if the associated attribute is a book, the ten backpacks may be arranged in context from the most suitable for carrying a book to the least suitable for carrying a book. Here, a backpack that frequently uses words such as “book”, “class”, and “school” in the text description will be placed in the most suitable for carrying the book. Any clusters (e.g., backpacks) associated with the object may be arranged and organized according to attributes (e.g., suitability for carrying a book) in a similar manner.

  In addition, the degree of association (eg, between objects and attributes, between objects and clusters) and the relative relationships between objects allows the search application to infer (as appropriate) aspects of one or more objects. It may be allowed to derive. For example, consider the case of a first cluster, a truffle and a second cluster, and a pet product. The fact that the text description of the object in the first cluster repeatedly contains the word “cook” and the lack of the word in the text description of the object in the second cluster indicates that the search application It is acceptable to determine that it is more related to gourmet cuisine. This reasoning can then be used to focus the search (eg gourmet cuisine). In fact, the degree of association described in this article may be used to retrieve various objects in a cluster based on non-traditional attributes. If appropriate, these relative relationships may be presented to the user when retrieving the search results as a knob that can be operated by the user. This is to give the user context on how related the returned results can be. Further, as will be discussed in more detail later, parameters (eg, attributes, degree of association and relative relationships) determined as discussed above may allow a user to dynamically adjust aspects of the search. It may be used to provide a search coordination mechanism.

  FIG. 5 illustrates groups and clusters of objects according to some embodiments. In this example, the primary cluster 100 may relate to “gourmet” objects that may be provided online by vendors, for example. Also, as discussed above, data associated with the objects may be analyzed to arrange the objects in the primary cluster 100.

  The illustrated primary cluster 100 includes three secondary clusters. The secondary cluster 101 may be related to “Truffle”, the secondary cluster 102 may be related to “Olive”, and the third cluster 103 may be related to “Cheese”. In this example, for the attribute “gourmet”, the arrangement of secondary clusters from most gourmet to least gourmet can be shown by arranging from left (ie most gourmet) to right (ie least gourmet).

  The illustrated secondary cluster 101 and truffle include four objects. Object 104 may be a sea salt truffle, object 105 may be a scorzone truffle, object 106 may be a bohemian truffle, and object 107 may be a summer banana split truffle. . Again in this example, with respect to the attribute “gourmet”, the arrangement of objects from the most gourmet to the least gourmet is shown by arranging from left (ie most gourmet) to right (ie least gourmet) in the secondary cluster. Can be done.

  FIG. 6 shows the degree of association between different products based on different attributes. For the objects 104-107 in FIG. 5, FIG. 6 shows that the objects can be arranged according to the first attribute 108 and the second attribute 109. In this example, the first attribute 108 may relate to “beach”, while the second attribute 109 may relate to “gourmet”. As shown, when arranged according to “gourmet”, sea salt truffles are the most gourmet, followed by scorzone truffles, bohemian truffles, and summer banana split truffles. On the other hand, when arranged according to “beach”, summer banana split truffles may be most suitable for a day at the beach, followed by sea salt truffles, bohemian truffles, and scorzone truffles. As discussed above, these relationships between objects may be derived, for example, from analysis of related text information from various online vendors. In fact, other unexpected attributes such as “breakfast” (eg breakfast preference) and “garden” (eg ease of growing in a home garden) may be used to characterize the relative relationships between objects. Good.

  FIGS. 7A-C illustrate a graphical user interface (GUI) 110 that may facilitate searching according to some embodiments. The GUI 110 may include a search bar 111, a result window 112, a first adjustment mechanism 113, a second adjustment mechanism 114, and a third adjustment mechanism.

  Search bar 111 may be used to enter search terms. For example, the user may enter the word “gourmet mushroom” as a search request. Here, the result window 112 may be used to display the search result. In the illustrated example, the search request returns “sea salt truffle” as the main result. As already mentioned, the user may determine that the sea salt truffle is not what he is looking for and thus use the adjustment mechanisms 113-115 to dynamically adapt the search. More specifically, the first adjustment mechanism 113, the second adjustment mechanism 114, and the third adjustment mechanism 115 may be graphical widgets that can be used to adjust the search. In this example, the adjustment mechanisms 113-115 may take the form of a rotatable knob. In another embodiment, the adjustment mechanisms 113-115 may take the form of a scrollable bar, for example.

  The first adjustment mechanism 113 may be related to the first aspect of the original search request, while the second adjustment mechanism 114 may be related to the second aspect of the search request. Thus, for example, the first adjustment mechanism 113 may be used to focus the resulting set of results more on the attribute gourmet, while the second adjustment mechanism 114 may be used to obtain the acquired results. May be used to focus more on attribute mushrooms. In other words, the user may increase the second adjustment mechanism 114 and / or decrease the first adjustment mechanism 113 to retrieve results that are more closely related to the mushroom than gourmet. Doing so would cause the search application to analyze the association information related to sea salt truffles to determine if there are any other objects that are related to gourmet but have a stronger relationship with the mushrooms. Good. FIG. 7B shows that in one example, the reconciliation process may result in a book titled “Anatomy of a Mushroom” (eg, this book treats the mushroom as delicious). May have chapters). Similarly, the user may reduce the second adjustment mechanism 114 and / or increase the first adjustment mechanism 113 to retrieve results that are more closely related to gourmet than mushrooms. Such an action may cause a search application to analyze association information related to sea salt truffles to determine if there are any objects related to mushrooms but having a stronger relationship with gourmets. .

  A third adjustment mechanism 115 may be used to adjust the search based on attributes associated with two or more objects in the object space. In this example, the third adjustment mechanism 115 may be used to adjust the influence of the beach on the search results. Thus, for example, when a user searches for gourmet mushrooms (ie, primary clusters) and the retrieved results relate to truffles (ie, secondary clusters) in general, the user (if desired) can direct the search to the beach. A third adjustment mechanism 115 may be used to further focus on the truffle for going out. At this point, the search application may analyze association information related to sea salt truffles to retrieve more focused results for truffles appropriate for the beach. FIG. 7C illustrates that in one example, the search application may return “Summer Banana Split Truffles” as the primary result in response to such adjustments. The user may use the third adjustment mechanism 115 in the reverse manner (to reduce rather than increase the beach influence in the search results). Of particular note is that the function of the third adjustment mechanism 115 may not include any search terms of the user's original search request. Indeed, it may be determined that “beach” is not a term used characteristically to quantify truffles, but can still be a useful term for end users in finding objects of interest.

  Turning now to FIG. 8, a method 600 for tracing data is shown. In this example, a user may utilize a GUI, such as a GUI (A-C in FIG. 7), to search for gourmet truffles to purchase from products provided by online vendors.

  The method 60 may be performed on a machine or computer readable storage medium such as a random access memory (RAM), read only memory (ROM), programmable ROM (PROM), flash memory, for example, a programmable logic array (PLA), In configurable logic such as field programmable gate arrays (FPGAs), composite programmable logic devices (CPLDs), for example application specific integrated circuits (ASICs), complementary metal oxide semiconductors (CMOS) or It may be implemented in a fixed function logic hardware using circuit technology such as transistor-transistor logic (TTL) technology or as a collection of logic and / or firmware instructions stored in any combination thereof. For example, computer program code for performing the operations shown in method 60 includes an object oriented programming language such as C ++ and a conventional procedural programming language such as C programming language or similar programming language. It may be written in any combination of one or more programming languages. The method may begin at process block 62.

  At processing block 64, the user may enter a term such as “gourmet mushroom” using a search bar such as search bar 111 (A-C in FIG. 7) to initiate the search. At processing block 66, the search application may associate the search term with a primary cluster, such as primary cluster 100 (FIG. 5), based on the “gourmet” attribute.

  At processing block 68, the search application may further focus the search to one or more secondary clusters to return search results. For example, a first secondary cluster, such as cluster 101 (FIG. 5), may relate to truffles. A second secondary cluster, such as secondary cluster 102 (FIG. 5), may relate to olives. At processing block 70, the search application may return results regarding the truffles.

  Along with the returned results, the search application may provide an adjustment mechanism for navigating the search results. Specifically, the search application analyzes various aspects of the cluster (eg, attributes, degree of association, etc.), and the first aspect of the user ’s search may be “gourmet” and the second aspect It may be determined that may be a “mushroom”. Thus, the search application relates to a first gourmet adjustment mechanism such as the first adjustment mechanism 115 (A to C in FIG. 7) and a mushroom such as the second adjustment mechanism 114 (A to C in FIG. 7). A second adjustment mechanism may also be provided. Furthermore, the search application can use a third adjustment mechanism similar to the third adjustment mechanism 115 (FIG. 7A-C) that can be used to adjust the search results based on attributes such as “beach”. May be included. In this example, the adjustment mechanism may take the form of a knob.

  At processing block 72, the user may review the results and use the first and second adjustment mechanisms to adjust the focus of the search to return more favorable results. Thus, in this example, the user may use a second adjustment mechanism to request that search results be more focused on mushrooms than gourmets.

  At processing block 74, the search application may utilize the user's input to recognize that the user prefers the mushroom. Thus, in this case, the search application analyzes the objects in the object space (using the degree of association and relative relationships discussed above), and the most appropriate result is a specific object such as the book “Mushroom Theory”. May be determined. At processing block 76, the search application may return the results to the user.

  At processing block 78, the user may view the results and determine that the book “Mushroom Theory” is too closely related to mushrooms and is not fully related to gourmet food suitable for the beach. Thus, at processing block 80, the user uses the first adjustment mechanism (related to gourmet) to adjust the search results focus back to gourmet food and uses the third adjustment mechanism to attribute. “Beach” may be incorporated into the process of tracing data.

  At processing block 82, the search application may analyze the objects in the first secondary cluster of the object space based on the user's request. That is, the search application (for example, using degree of association, relative relationships, etc.) is more related to gourmet food than summer banana split truffles than other objects in the object map, It may be determined that they are related. At processing block 84, the search application may return a second result to the user. At processing block 86, the user may conduct a transaction to purchase a summer banana split truffle from an online vendor offering it. In process block 88, the process may end. The ordering and numbering of the blocks depicted in FIG. 8 is not intended to imply an order of operation, excluding other possibilities. Those skilled in the art will appreciate that the systems and methods described above can be subject to various modifications, variations and modifications.

  For example, in the embodiment shown in FIG. 8, the object may be parsed according to attributes related to the object. More specifically, the results to be returned to the user have been analyzed with respect to the truffle object's beach attribute, but the search application analysis need not be limited to the attribute with respect to the object. In other embodiments, the search application may take other information into account. Thus, for example, the search application may analyze the user profile to determine that the user is aware of environmental issues. In this case, the search application may take this aspect into account by providing an adjustment mechanism for adjusting to prioritize environmentally friendly products.

  FIG. 9 illustrates a processor core 200 according to an embodiment. The processor core 200 may be a core for any type of processor such as a microprocessor, embedded processor, digital signal processor (DSP), network processor or other device executing code. Although only one processor core 200 is shown in FIG. 9, the processing elements may alternatively include two or more of the processor cores 200 shown in FIG. The processor core 200 may be a single thread core, or for at least one embodiment, the processor core 200 may have more than one hardware thread context (or “logical processor”). It may be multithreaded in that it can be included.

  FIG. 9 also shows a memory 270 coupled to the processor 200. The memory 270 may be any of a wide variety of memories (including various layers of the memory hierarchy) that are known or otherwise available to those skilled in the art. Memory 270 may include one or more code 213 instructions executed by processor 200 core. Code 213 may implement logical architecture 36 (FIG. 2) already discussed. The processor core 200 follows the program sequence of instructions indicated by code 213. Each instruction enters the front end portion 210 and may be processed by one or more decoders 220. The decoder 220 may generate a micro-operation at its output, such as a fixed-width micro-operation in a predefined format, or other instruction, micro-instruction or control signal that reflects the original code instruction. It may be generated. The illustrated front end 210 also includes register rename logic 225 and scheduling logic 230. These generally allocate resources and queue the operations corresponding to the conversion instructions for execution.

  The processor 200 is shown as including execution logic 250 having a set of execution units 225-1 through 225-N. Some embodiments may include several execution units dedicated to specific functions or sets of functions. Other embodiments may include only one execution unit or one execution unit that can perform a particular function. The illustrated execution logic 250 performs the operation specified by the code instruction.

  After completing the execution of the operation specified by the code instruction, the back-end logic 260 retires the code 213 instruction. In some embodiments, the processor 200 allows out-of-order execution, but requires in-order retirement of instructions. The rear logic 265 can take a variety of forms known to those skilled in the art (eg, reordering buffers, etc.). In this way, processor core 200 is modified by at least the output generated by the decoder, hardware registers and tables utilized by register rename logic 225 and execution logic 250 during execution of code 213. Conversion is performed at an arbitrary register point (not shown).

  Although not shown in FIG. 9, the processing elements may include a processor core 200 and other elements that are on-chip. For example, the processing element may include memory control logic along with the processor core 200. The processing element may include I / O control logic and / or may include I / O control logic integrated with memory control logic. The processing element may include one or more caches.

  Referring now to FIG. 10, a block diagram of a system embodiment 1000 according to an embodiment of the present invention is shown. Shown in FIG. 10 is a microprocessor system 1000 that includes a first processing element 1070 and a second processing element 1080. Although two processing elements 1070 and 1080 are shown, it is understood that embodiments of the system 1000 may include only one such processing element.

  System 1000 is shown as a point-to-point interconnect system, where first processing element 1070 and second processing element 1080 are coupled via point-to-point interconnect 1050. It should be understood that any or all of the interconnections shown in FIG. 10 may be implemented as a multi-drop bus rather than a point-to-point interconnection.

  As shown in FIG. 10, each of processing elements 1070 and 1080 are multi-core processors including first and second processor cores (ie, processor cores 1074a and 1074b and processor cores 1084a and 1084b). May be. Such cores 1074, 1074b, 1084a, 1084b may be configured to execute instruction code in a manner similar to that discussed above in connection with FIG.

  Each processing element 1070, 1080 may include at least one shared cache 1896. Shared caches 1896a, 1896b may store data (eg, instructions) utilized by one or more components of the processor, such as cores 1074a, 1074b and 1084a, 1084b, respectively. For example, a shared cache may cache data stored in memory 1032, 1034 locally for faster access by processor components. In one or more embodiments, the shared cache is one or more intermediate level caches such as level 2 (L2), level 3 (L3), level 4 (L4) or other level caches. , Last level cache (LLC) and / or combinations thereof.

  Although only two processing elements 1070, 1080 are shown, it should be understood that the scope of the present invention is not limited thereto. In other embodiments, one or more additional processing elements may be present in a given processor. Alternatively, one or more of the processing elements 1070, 1080 may be an element other than a processor, such as an accelerator or a field programmable gate array. For example, the additional processing element may be the same additional processor (s) as the first processor 1070, the additional processor (s) heterogeneous or asymmetric with the first processor 1070, the accelerator (E.g., a graphics accelerator or digital signal processing (DSP) unit), a field programmable gate array, or any other processing element. There may be a variety of differences between the processing elements 1070, 1080 in terms of a series of indicators including structure, microstructure, heat, power consumption characteristics, and the like. These differences can effectively surface asymmetry and heterogeneity between the processing elements 1070, 1080. For at least one embodiment, the various processing elements 1070, 1080 may be in the same die package.

  The first processing element 1070 may further include memory controller logic (MC) 1072 and point-to-point (P-P) interfaces 1076 and 1078. Similarly, the second processing element 1080 may include MC 1082 and PP interfaces 1086 and 1088. As shown in FIG. 10, MCs 1072 and 1082 couple the processor to respective memories, namely memory 1032 and memory 1034. These memories may be part of main memory that is locally attached to the respective processor. Although MC logic 1072 and 1082 are shown as being integrated into processing elements 1070, 1080, for alternative embodiments, MC logic is not integrated within processing elements 1070, 1080; There may be separate logic outside the processing elements 1070, 1080.

  First processing element 1070 and second processing element 1080 may be coupled to I / O subsystem 1090 via PP interconnections 1076, 1086, and 1084, respectively. As shown in FIG. 10, I / O subsystem 1090 includes P-P interfaces 1094 and 1098. Further, the I / O subsystem 1090 includes an interface 1092 that couples the I / O subsystem 1090 with the high performance graphics engine 1038. In some embodiments, bus 1049 may be used to couple graphics engine 1038 to I / O subsystem 1090. Alternatively, a point-to-point interconnect 1039 may couple these components.

  In turn, the I / O subsystem 1090 may be coupled to the first bus 1016 via the interface 1096. In some embodiments, the first bus 1016 may be a peripheral component interconnect (PCI) bus or a bus such as PCI Express or other third generation I / O interconnect bus. However, the scope of the present invention is not limited thereto.

  As shown in FIG. 10, various I / O devices 1014 may be coupled to the first bus 1016 along with a bus bridge 1018 that may couple the first bus 1016 to the second bus 1010. In some embodiments, the second bus 1020 may be a low pin count (LPC) bus. Various devices may be coupled to the second bus 1020. Such various devices include, for example, keyboard / mouse 1012, network controller (s) / communication device (s) 1026 (which may be in communication with computer network 503) and in certain embodiments. A data storage unit 1018 such as a disk drive or other mass storage device that may include code 1030 is included. The communication device 1026 may be used to receive a text description of the objects already discussed. Code 1030 may include instructions for performing an embodiment of one or more of the above methods. Thus, the illustrated code 1030 may implement the logical architecture 36 (FIG. 2) and may be similar to the code 213 (FIG. 9) already discussed. Further, an audio I / O 1024 may be coupled to the second bus 1020.

  Note that other embodiments are contemplated. For example, instead of the point-to-point architecture of FIG. 10, the system may implement a multi-drop bus or other such communication technology. Also, the elements of FIG. 10 may alternatively be divided using more or fewer integrated chips than those shown in FIG.

  Thus, an example may include an object space generation system having a network controller that receives a plurality of text descriptions. Here, each text description includes one or more words and phrases that describe the object of the plurality of objects. The system further includes a first association module for determining a first degree of association between each of the plurality of text descriptions and the one or more topics to obtain first association information; A second association module for determining a second degree of association between each of the plurality of objects and the one or more topics to obtain second association information. Good. Furthermore, the system may include a cluster module that groups the plurality of objects into clusters based on the first association information and the second association information.

  In addition, the system may further include a topic generation module that identifies the one or more topics based on a plurality of text descriptions.

  Further, the topic generation module may use a topic module to identify the one or more topics.

  Furthermore, the first association module may configure the first association information as a first set of vectors corresponding to the plurality of objects, and the second association module may include the second association module. The association information may be configured as a set of vectors corresponding to the plurality of text descriptions.

  In addition, the system further provides a third association that determines a third degree of association between each text description and one or more words in the text description to obtain third association information. A module and a fourth association module that determines a fourth degree of association between each word and the one or more topics to obtain fourth association information. The first association module may obtain the first association information using the third and fourth association information, and the second association module may obtain the third and fourth association information. It may be used to obtain the fourth association information.

  Further, the first association module may use a probabilistic graphical model (PGM) to propagate the fourth association information to the plurality of text descriptions.

  Further, the second association module may use a probabilistic graphical model (PGM) to propagate the fourth association information to the plurality of objects.

  Furthermore, the third association module may configure the third association information as a third set of vectors corresponding to the plurality of text descriptions, and the fourth association module may include the fourth association module. May be configured as a fourth set of vectors corresponding to the one or more words.

  Further, each vector in the third set of vectors may have a weight value that identifies the occurrence frequency of the one or more words in the respective text description.

  Further, the cluster module may spatially order the clusters based on the relative relationships between objects and clusters.

  In another example, a method for generating an object space includes determining a first degree of association between each of a plurality of text descriptions and one or more topics to obtain first association information; Determining a second degree of association between each of the plurality of objects and the one or more topics to obtain second association information; the first association information and the second Grouping the plurality of objects into clusters based on association information.

  In addition, each text description may include one or more of words and phrases that describe an object of the plurality of objects. Here, the method further includes identifying the one or more topics based on the plurality of text descriptions.

  In addition, the method further includes using a topic model to identify the one or more topics.

  In addition, the method further comprises configuring the first association information as a first set of vectors corresponding to the plurality of objects, and the second association information corresponding to the plurality of text descriptions. It may include constructing as two sets of vectors.

  In addition, determining the first degree of association and the second degree of association includes determining a third degree of association between each text description and one or more words in the text description. Determining to obtain third association information, and determining a fourth degree of association between each word and the one or more topics to obtain fourth association information. .

  In addition, the method may further include using a probabilistic graphical model (PGM) to propagate the fourth association information to the plurality of text descriptions.

  In addition, the method may further include using a probabilistic graphical model (PGM) to propagate the fourth association information to the plurality of objects.

  In addition, the method further comprises configuring the third association information as a third set of vectors corresponding to the plurality of text descriptions, and converting the fourth association information into the one or more words. Constructing as a corresponding fourth set of vectors may be included.

  Further, each vector in the third set of vectors may have a weight value that identifies the occurrence frequency of the one or more words in the respective text description.

  Further, grouping the plurality of objects may include spatially ordering the clusters based on a relative relationship between the objects.

Other examples include at least a set of traversal instructions that, when executed by a processor, cause the computer to generate a first set of results based on the requested search scope and object space. One computer-readable storage medium may be included. Here, the object space includes a first object and a second object arranged according to a relative relationship based on a certain attribute. When the follow instruction is executed, it is further sent to the computer.
Generating a user interface including a first adjustment mechanism configured to adjust a requested search scope and a second adjustment mechanism configured to adjust a search result based on the attribute; Generating a second set of results based on user input received via one or more of the first adjustment mechanism and the second adjustment mechanism, wherein the second set of results is A step including the second object may be performed.

  Further, the attribute of the medium may be based on words included in a first text description of the first object and a second text description of the second object.

  Further, the user interface of the medium may be a graphical user interface (GUI) and the first adjustment mechanism may be one of a knob and a scrollable bar.

  Further, the user interface of the medium may be a graphical user interface (GUI) and the second adjustment mechanism may be one of a knob and a scrollable bar.

  The example may also include a data traversing device that includes a first result module that generates a first set of results based on a requested search scope and object space. Here, the object space includes a first object and a second object arranged according to a relative relationship based on a certain attribute. The data tracing device includes a first adjustment mechanism configured to adjust a requested search range and a second adjustment mechanism configured to adjust a search result based on the attribute. An adjustment module for generating an interface and a second result for generating a second set of results based on user input received via one or more of the first adjustment mechanism and the second adjustment mechanism And the second set of results includes the second object.

  Further, the attribute of the device may be based on words contained in a first text description of the first object and a second text description of the second object.

  Further, the user interface of the device may be a graphical user interface (GUI), and the first adjustment mechanism is one of a knob and a scrollable bar.

  Further, the user interface of the device may be a graphical user interface (GUI) and the second adjustment mechanism is one of a knob and a scrollable bar.

  Various embodiments may be implemented using hardware elements, software elements, or a combination of both. Examples of hardware elements are processors, microprocessors, circuits, circuit elements (eg, transistors, resistors, capacitors, inductors, etc.), integrated circuits, application specific integrated circuits (ASICs), programmable logic devices (PLDs) , Digital signal processors (DSPs), field programmable gate arrays (FPGAs), logic gates, registers, semiconductor devices, chips, microchips, chipsets, and the like. Examples of software include software components, programs, applications, computer programs, application programs, system programs, machine programs, operating system software, middleware, firmware, software modules, routines, subroutines, functions, methods, Includes procedures, software interfaces, application program interfaces (APIs), instruction sets, computing code, computer code, code segments, computer code segments, words, values, symbols, or any combination thereof You may go out. The determination of whether an embodiment is implemented using hardware and / or software elements depends on the desired calculation rate, power level, thermal tolerance, processing cycle budget, input data rate, output data rate , Memory resources, data bus speed, and other design or performance constraints can vary.

  One or more aspects of at least one embodiment may be implemented by representing instructions stored on a machine-readable medium that represents various logic within the processor. The instructions, when read by a machine, cause the machine to produce logic that performs the techniques described herein. Such representations, known as “IP cores”, are stored on tangible machine-readable media and supplied to various customers or manufacturing facilities for loading into the production machine that actually forms the logic or processor. Also good.

  Embodiments of the present invention are applicable to any type of semiconductor integrated circuit (“IC”) chip. Examples of these IC chips include but are not limited to processors, controllers, chipset components, programmable logic arrays (PLAs), memory chips, network chips, and the like. Further, in some of the drawings, the signal conducting lines are represented by lines. Some are different to indicate more component signal paths, have numeric labels to indicate the number of component signal paths, and / or indicate the direction of the main information flow May have arrows at one or more ends, but this should not be construed in a limiting manner. Rather, such additional details may be used in connection with one or more exemplary embodiments to facilitate easier understanding of the circuit. Any signal line represented, whether or not it has additional information, may actually have one or more signals that can travel in multiple directions, and any suitable type. For example, digital or analog lines implemented with differential pairs, fiber optic lines and / or single-ended lines.

  Although exemplary sizes / models / values / ranges may be given, embodiments of the invention are not so limited. As manufacturing techniques (eg, photolithography) mature over time, it is expected that smaller sized devices can be manufactured. Further, well-known power / ground connections to the IC chip and other components are included in the drawings for ease of illustration and discussion, and to avoid burying certain aspects of embodiments of the present invention. It may or may not have been shown. Further, arrangements may be shown in block diagram form to avoid burying embodiments of the present invention. This also takes into account the fact that details regarding the implementation of such a block diagram arrangement are highly dependent on the platform on which the embodiments are implemented, i.e. such details should be well within the scope of those skilled in the art. It is also a thing. Where specific details (eg, circuitry) are described to describe exemplary embodiments of the invention, those skilled in the art will recognize that embodiments of the invention may be without or without those specific details. It should be apparent that this can be done with variations of individual details. Thus, the description is to be regarded as illustrative rather than limiting.

  Some embodiments, for example, a machine or tangible machine may store instructions or a set of instructions that, when executed by a machine, may cause the machine to perform the methods and / or operations according to the above embodiments. It may be implemented using a computer readable medium or article. Such machines may include, for example, any suitable processing platform, computing platform, computing device, processing device, computing system, processing system, computer, processor, etc., hardware and / or Or it may be implemented using any suitable combination of software. The machine-readable medium or article may be, for example, any suitable type of memory unit, memory device, memory article, memory medium, storage device, storage article, storage medium and / or storage unit, such as memory, removable or non-removable Removable media, erasable or non-erasable media, writable or rewritable media, digital or analog media, hard disk, floppy disk, compact disk read only memory (CD-ROM), recordable compact disk (CD -R), rewritable compact discs (CD-RW), optical discs, magnetic media, magneto-optical media, removable memory cards or discs, various types of digital versatile discs (DVDs), tapes, cassettes, etc. You may go out. The instructions can be any suitable high-level, low-level, such as source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, etc. It may include any suitable type of code implemented using a programming language that is level, object-oriented, visual, compiled and / or interpreted.

  Unless stated otherwise, terms such as “processing”, “computing”, “calculation”, “decision”, etc. refer to the physical (eg, electronic) of a computer or computing system or similar electronic computing device. Manipulate and / or convert data represented in the registers and / or memory of the computing system as a significant quantity within the memory, registers or other such information storage, transmission or display device of the computing system It can be understood that it refers to other data operations and / or processes that are similarly expressed as physical quantities. Embodiments are not limited in this respect.

  The term “coupled” is used herein to refer to any type of direct or indirect relationship between the components in question: electrical, mechanical, fluid, optical It can be applied to electrical, electromagnetic, electromechanical or other connections. Furthermore, terms such as “first” and “second” may only be used in this article to facilitate discussion, and unless otherwise noted, have specific temporal or chronological significance. Sometimes it does n’t.

  Those skilled in the art will appreciate from the foregoing description that the broad techniques of the embodiments of the present invention can be implemented in a variety of forms. Thus, although embodiments of the present invention have been described in the context of specific examples thereof, the true scope of embodiments of the present invention should not be limited thereto. Other modifications will become apparent to those skilled in the art after reviewing the drawings, specification, and appended claims.

Claims (20)

An object space generation system:
A network controller that receives a plurality of text descriptions, each text description including one or more words and phrases that describe objects of the plurality of objects; and
A first association module that determines a first degree of association between each of the plurality of text descriptions and one or more topics to obtain first association information;
A second association module that determines a second degree of association between each of the plurality of objects and the one or more topics to obtain second association information;
A cluster module that groups the plurality of objects into clusters based on the first association information and the second association information;
system.   The system of claim 1, further comprising a topic generation module that identifies the one or more topics based on the plurality of text descriptions.   The system of claim 2, wherein the topic generation module uses a topic model to identify the one or more topics.   The first association module configures the first association information as a first set of vectors corresponding to the plurality of text descriptions, and the second association module includes the second association information, The system according to claim 1, wherein the system is configured as a set of vectors corresponding to the plurality of objects. A third association module that determines a third degree of association between each text description and one or more words in the text description to obtain third association information;
A fourth association module that determines a fourth degree of association between each word and the one or more topics to obtain fourth association information;
The first association module obtains the first association information using the third and fourth association information, and the second association module uses the third and fourth association information to obtain the first association information. The system according to claim 1, wherein the fourth association information is obtained.   6. The system of claim 5, wherein the first association module uses a probabilistic graphical model (PGM) to propagate the fourth association information to the plurality of text descriptions.   6. The system of claim 5, wherein the second association module uses a probabilistic graphical model (PGM) to propagate the fourth association information to the plurality of objects.   The third association module configures the third association information as a third set of vectors corresponding to the plurality of text descriptions, and the fourth association module comprises the fourth association information, 6. The system of claim 5, configured as a fourth set of vectors corresponding to the one or more words.   9. The system of claim 8, wherein each vector in the third set of vectors has a weight value that identifies the frequency of occurrence of the one or more words in a respective text description.   10. A system according to any one of the preceding claims, wherein the cluster module spatially orders the clusters based on a relative relationship between the objects and clusters. A method of creating an object space that:
Determining a first degree of association between each of the plurality of text descriptions and one or more topics to obtain first association information;
Determining a second degree of association between each of a plurality of objects and the one or more topics to obtain second association information;
Grouping the plurality of objects into clusters based on the first association information and the second association information,
Method.   Each text description includes one or more of words and phrases describing an object of the plurality of objects, and the method further identifies the one or more topics based on the plurality of text descriptions 12. The method of claim 11, comprising:   The method of claim 12, further comprising using a topic model to identify the one or more topics. Configuring the first association information as a first set of vectors corresponding to the plurality of objects;
Further comprising configuring the second association information as a second set of vectors corresponding to the plurality of text descriptions;
14. A method according to any one of claims 11 to 13. At least one computer readable storage medium that includes a set of instructions to be traced to a computer when executed by a processor:
Generating a first set of results based on a requested search scope and object space, wherein the object space comprises a first object and a second object arranged according to a relative relationship based on a certain attribute; Including a stage;
Generating a user interface including a first adjustment mechanism configured to adjust the scope of the requested search and a second adjustment mechanism configured to adjust a search result based on the attribute; ;
Generating a second set of results based on user input received via one or more of the first adjustment mechanism and the second adjustment mechanism, the second set of results Includes performing the steps including the second object,
Computer-readable storage medium.   The at least one computer-readable medium of claim 15, wherein the attribute is based on words included in a first text description of the first object and a second text description of the second object.   17. At least one computer readable medium according to claim 15 or 16, wherein the user interface is a graphical user interface (GUI) and the first adjustment mechanism is one of a knob and a scrollable bar.   17. At least one computer readable medium according to claim 15 or 16, wherein the user interface is a graphical user interface (GUI) and the second adjustment mechanism is one of a knob and a scrollable bar. A data tracking device:
A first result module that generates a first set of results based on a requested search scope and object space, the object space comprising: a first object arranged according to a relative relationship based on an attribute; A first result module containing a second object;
An adjustment module for generating a user interface including a first adjustment mechanism configured to adjust the scope of the requested search and a second adjustment mechanism configured to adjust a search result based on the attribute When;
A second result module that generates a second set of results based on user input received via one or more of the first adjustment mechanism and the second adjustment mechanism; The second set of results includes the second object;
apparatus.   The apparatus of claim 19, wherein the attribute is based on words included in a first text description of the first object and a second text description of the second object.

Images